MasterNanosciences, nanotechnologies

This course is made of 2 subcourses of 3 credits each, which can be taken separately.

Statistical Physics I: Theory (3 ECTS)

Goal: Introduce the basic statistical physics concepts to address the equilibrium and evolution properties of nano-scale systems.

Content:

• Statistical entropy
• Boltzmann factor, statistics in the canonical ensemble, mean values and fluctuations
• classical and quantum oscillators, applications
• chemical potential, grand canonical ensemble,applications: adsorption isotherms, quantum statistics
• fluctuations dynamics, linear response
• Monte-Carlo notions

Statistical Physics II:  Introduction to Machine Learning for Physics (3 ECTS)

Goal: This course provides an introduction to machine learning and its applications in physics research. Machine learning and artificial intelligence are having an increasingly significant impact on scientific research and society as a whole. One of the main drivers of recent progress in these fields is the development of neural networks, particularly deep learning. The course is designed for M1 students in the physical sciences who have little or no previous experience with machine learning. It will introduce the basic principles of machine learning, neural networks, and deep-learning methods. The lectures will be complemented by hands-on tutorials and exercises to consolidate students’ understanding and provide practical experience with the methods discussed in class.
 
Content:
 

• Overview of machine learning: Supervised and unsupervised learning, regression, classification, generative models, training and test data, overfitting, and regularization.
• Neural networks: Perceptrons, activation and loss functions, gradient descent, multilayer neural networks, and backpropagation.
• Deep learning: Training deep neural networks, optimization, regularization, and convolutional neural networks (CNNs).
• Practical implimentation of simple CNN, filters and weights optimisation, and recovery of spatial information out of abstract feature channels.
• Applications to physics research: Examples of how machine-learning methods can be used in data analysis such as image classification and restoration.

The course includes lectures, hands-on tutorials, and exercises. Students are encouraged to bring their own laptop and use Google Colaboratory during the practical sessions. Knowledge of Python is required. For the introduction to neural networks and deep learning, the course will follow the online textbook Neural Networks and Deep Learning by Michael Nielsen: http://neuralnetworksanddeeplearning.com/